You show your ignorance.
FICS has a very large searchable database of every game played on the site. Very few people even bother to look at it because the vast majority of those games are worthless, from a competitive perspective and even from a theoretical perspective.
If a game analysis of a game between two patzers shows that a mate was misssed, it's not very useful for trying to solve chess as a whole.
Btw, this is why you don't have to map out every single move to claim you have solved chess.
Once a position is demonstrably won, it doesn't need to be played out. We can save several orders of magnitudes of moves by eliminating lost positions from the database of all games.
If we stipulate that a position is won or lost with best play, then the search for a solution tracks back down the variation to see whether an improvement can be made.
Basically, 1.f4 e6 2.g4 is lost, so that position doesn't need to be saved. Just a symbol is enough.
Same goes for all positions after 1.e4 e5 2.Qh5 Nf6 3.Qxf7, and so on.
Except that "saving several orders of magnitude" only drops the problem from 10^46.7 to, say, 10^44. Not really helping. Saying that you can prune/eliminate 999 out of every thousand positions without any evaluation taking place (which is BS, but let's say you could) still leaves a completely insurmountable amount of positions.
This talk of every possible position is nonsense. Fruitless. A complete waste of electricity. The search for AI is on, but it has nothing to do with brute force. It has been pointed out enough times here, a table base of every possible game leads absolutely nowhere in a search to "solve chess".
To that note; eliminate all possible positions by a 1000 fold, and you're still at ground zero.
Nowhere
If a solution exists, it rests well outside bits of 1's and O's.
Appears you think the solution lies in "numbers". Brute calculation.
I suggest this lies very far from a solution.
Appears you think the solution lies in "numbers". Brute calculation.
I suggest this lies very far from a solution.
There is no indication of even the possibility of a solution that does not use a huge amount of brute force, analogous to the solution of checkers, but so much larger it is infeasible with conventional computing technology. We have discussed some ways in which the amount of brute force can be reduced significantly, but it leaves a heck of a lot.
The solution of chess is a combinatorial problem, very much like a mathematical theorem with very arbitrary assumptions (the rules of chess). It is the arbitrariness of the rules of chess that offers no hint of shortcuts.
This seems a rather reliable belief. Consider a tiny version of the problem, say K+Q versus K+P endings. How could you conceivably determine whether a specific position was winning without analysing EVERY position the defense can reach against some selected choices for the side with the queen. Applied recursively, this is how a tablebase is created).
@Elroch
"It is the arbitrariness of the rules of chess that offers no hint of shortcuts."
Finally there is at least one person that can see things very clearly.
Consider en passant, you would not believe how such a little thing can complicate or affect so much code in chess engines, the reason? because it is not a standard move, 3 squares are affected and you have to maintain state using a global variable at all times in the program.
Then consider stalemate as a draw, a rule that only became a standard in the 19th century, it complicates chess so much it is a joke.
K + 2 Knights versus K is a draw but add one or more black pawns and it may be a win for White because Black can not rely on stalemate, it gets even more complicated, the first Troitsky Line in this ending often requires more that 50 moves to force checkmate so they had to create a second Troitsky Line for wins that are possible in 50 moves or less.
The winning procedure is so complicated and because it occurs very rarely in practice most grandmasters can not win the game even when it is possible.
Because of all this you need tablebases so that the engine can peek over the game horizon (it's deepest search depth) when calculating.
The file sizes of tablebases increases exponentially as the piece count increases, a 16 man tablebase, which does not yet exist, will probably not even fit on your hard drive no matter how big it is and in any case most moves in any existing tablebase do not even make sense to humans because the programs work backwards from every checkmate position.
Try it, find an online tablebase, throw a K + N + B v K on the board and try and make sense of the moves.
Then throw a K + N + N v K + P and see how that goes.
I made a game and I changed some rules and I did it for very specific reasons.
Good luck with the math, you gonna need a real big blackboard and lots of chalk.
In theory it should be the same as in practice, but in practice it is never the same as in theory.
lol
Don't get it. I agree with folks who claim brute force can never "solve chess", then those very same people state brute force, table bases is they only approach to the problem. Contradictions abound. People don't take the time to actually "read", quickly jumping into expressing their particular view, entirely missing points made by others.
During the games, AlphaGo played a handful of highly inventive winning moves, several of which - including move 37 in game two - were so surprising they overturned hundreds of years of received wisdom, and have since been examined extensively by players of all levels. In the course of winning, AlphaGo somehow taught the world completely new knowledge about perhaps the most studied and contemplated game in history.
One of the great promises of AI is its potential to help us unearth new knowledge in complex domains. We’ve already seen exciting glimpses of this, when our algorithms found ways to dramatically improve energy use in data centres - as well as of course with our program AlphaGo.
Yes, it is too complicated for me to answer.
But I can say that a computer does not have to tabulate every position individually. Logical conditions can be used to simplify the analysis and shorten the computation. For example if White's king is not in danger and Black's king is not stalemated and pawns are not blockaded and White has a big material advantage then White will win. Human minds and hands are not well equipped to define and understand these ideas exactly and completely but computers are much better at dealing with logical expressions, and they will be able to program themselves to accelerate the chess solution tremendously.